• No results found

Reference model for adaptive and intelligent educational systems supported by learning objects

N/A
N/A
Protected

Academic year: 2021

Share "Reference model for adaptive and intelligent educational systems supported by learning objects"

Copied!
141
0
0

Loading.... (view fulltext now)

Full text

(1)

Reference Model for Adaptive and

Intelligent Educational Systems

supported by Learning Objects

Julián Moreno Cadavid

Universidad Nacional de Colombia

Facultad de Minas, Departamento de Ciencias de la Computación y la Decisión Medellín, Colombia

(2)
(3)

Reference Model for Adaptive and

Intelligent Educational Systems

supported by Learning Objects

Julián Moreno Cadavid

Doctoral thesis submitted as partial requirement for the degree of:

Doctor in Engineering - Systems

Supervisor:

Demetrio Arturo Ovalle Carranza, Ph.D., Titular professor Research and Development Group in Artificial Intelligence

Doctoral committee: Magda Bercht, Ph.D.

Universidade Federal do Rio Grande do Sul, Brazil

Ricardo Azambuja Silveira, Ph.D.

Universidade Federal de Santa Catarina, Brazil

Néstor Darío Duque, Ph.D.

Universidad Nacional de Colombia - Manizales, Colombia

Universidad Nacional de Colombia

Facultad de Minas, Departamento de Ciencias de la Computación y la Decisión Medellín, Colombia

(4)
(5)

AKNOWLEDGMENTS

There are many people who have helped me to fulfill the research presented on this document. They have all, in their own way, made this achievement possible. In particular, I would like to present my gratitude to the following individuals.

First, I thank my advisors. I was fortunate of having not just one but two of them. To Professor Demetrio Arturo Ovalle, PhD, who has guided and supported me during my entire academic life. To Professor Rosa Maria Vicari, PhD, who always receive me with warm hospitality and showed me the value of helping others. Far beyond this thesis, they both have been a reference point for me.

Second, I thank the rest of the committee members for accepting such a task and for providing extremely valuable comments. To Professor Ricardo Azambuja Silveira, PhD, who kindly invited me at the Universidade Federal de Santa Catarina. To Professor Magda Bercht, PhD, from who I have the pleasure of receiving a course during my studies at the

Universidade Federal de Rio Grande do Sul. To Professor Nestor Darío Duque, PhD who has been a remarkable example not only with his own dissertation, but as researcher and as person.

Third, I thank the students and teacher of the post-graduate program Maestría en Enseñanza de las Ciencias Exactas y Naturales at the Universidad Nacional de Colombia – Medellín who attended the course Taller TICs y Educación en Ciencias I during semester 2012-1. All you guys were an enormous help in the last parts of this research.

I also express my infinite gratitude to my beloved family: my wife, my mother and my sister. All I am is because of you and all I do is because of you.

Finally, I would like to thank the Instituto Colombiano para el Desarrollo de la Ciencia y la Tecnología “Francisco José de Caldas” - COLCIENCIAS, which sponsored my doctoral studies.

(6)
(7)

Reference model for intelligent and adaptive educational systems

supported by learning objects

ABSTRACT

Computer Aided Learning, known more widely with the generic name of e-learning, has become a powerful tool with lots of potentialities within educational field. Even though, one of the main critics that it receives is that in most cases the implemented courses follows a “one size fits all” approach, which means that all students receive the same content in the same way being unaware of their particular needs. This problem is not due only to the absence of direct interaction between student and tutor, but also because of the lack of an appropriate instructional design.

There are several approaches which deal with this issue and look for adapt the teaching process to students. One could say that in the top of those approaches the Adaptive and Intelligent Educational Systems are situated, which merges the functionalities of two approaches: the Adaptive Educational Hypermedia Systems and the Intelligent Tutoring Systems. Nevertheless, after an extensive literature review, a major inconvenience is still found for this kind of systems and particularly for their reference models: or they are too simple, including just a few functionalities; or they are too complex, which difficult their design and implementation. Considering this panorama, the main objective of this dissertation thesis was the definition of a reference model trying to reach such an elusive equilibrium, in such a way that allows the design of courses which adapt themselves in an intelligent and effective way to the progress and characteristics of each student but without being too complex. Another important feature is that this model integrates Learning Objects, promoting this way flexibility and reusability.

In order to achieve this general objective, three sub-models were considered: a domain model, a student model and a tutor model. The first one serves to structure the knowledge domain and was defined using the notion of learning goal and a flexible multilevel schema with optional prerequisite operations. The second one aids to characterize students and considered personal, knowledge and psycho-cognitive information. The third one may be considered as the hearth of the system and defines the adopted adaptive functionalities: sequencing and navigation, content presentation, assessment, and collaborative support.

With the aim of clarify the three sub-models, as well as all their components and relationships, an instantiation example was also presented. Such an instantiation was called Doctus, an authoring tool for adaptive courses. Doctus was not only helpful to exemplify the setup of the referece model as a whole, but also to refine sub-models and several procedures envolved. As final part of the dissertation, the implementation and preliminary validation of Doctus was performed. This was done with 51 subjects, teachers from

(8)

different formation levels. The obtained results in this stage were outstanding, all the adaptive functionalities were well evaluated and all of those polled felt enthusiastic about counting with a tool for helping them in their teaching practices considering students as particular individuals.

(9)

Modelo de referencia para sistemas educacionales adaptativos inteligentes

soportados por objetos de aprendizaje

RESUMEN

El aprendizaje asistido por computador, conocido más ampliamente con el nombre genérico de e-learning, se ha convertido en una poderosa herramienta con amplias potencialidades dentro del campo educativo. Aun así, una de las mayores críticas que este recibe es que en la mayoría de los casos los cursos que son implementados siguen un enfoque “one size fits all”, es decir, que todos los alumnos reciben exactamente el mismo

contenido y de la misma manera desconociendo sus necesidades particulares. Esta falla radica no sólo en la falta de interacción directa entre alumno y tutor, sino también en la falta de un diseño instruccional apropiado que considere diversos de los enfoques disponibles hoy en día.

Existen diversos enfoques que buscan solucionar este problema y adaptar el proceso de enseñanza a los estudiantes. Se podría decir que a la vanguardia de estos enfoques se encuentran los Sistemas Educacionales Inteligentes Adaptativos, los cuales combinan las funcionalidades de dos enfoques: los Sistemas Hipermedia Educacionales Adaptativos y los Sistemas Tutoriales Inteligentes. Sin embargo, luego de una extensa revisión bibliográfica, se encontró que existe aún un inconveniente importante con este tipo de sistemas y en particular con sus modelos de referencia: o son demasiado simples, incluyendo solamente unas pocas funcionalidades; o son demasiado complejos, lo cual dificulta su diseño e implementación. Considerando este panorama, el objetivo principal de esta tesis fue la definición de un modelo de referencia intentando alcanzar tal equilibrio esquivo, de tal manera que permita el diseño de cursos que se adapten de una manera efectiva e inteligente al progreso y características de cada estudiante pero sin ser demasiado complejo. Otra propiedad importante de dicho modelo es que integra el uso de Objetos de Aprendizaje, promoviendo así la flexibilidad y la reusabilidad.

Con el fin de alcanzar este objetivo general, tres sub modelos fueron considerados: un modelo del dominio, un modelo del estudiante y un modelo del tutor. El primero sirve para estructurar el dominio de conocimiento y fue definido empleando la noción de objetivo de aprendizaje junto con un esquema flexible multinivel con operaciones opcionales de prerrequisitos. El segundo busca caracterizar los estudiantes y considera información personal, de conocimiento y psico-cognitiva. El tercero puede ser considerado como el corazón del sistema y define las funcionalidades adaptativas consideradas: secuenciamiento y navegación, presentación de contenido, evaluación, y soporte colaborativo.

Con el fin de clarificar los tres sub modelos, así como todos sus componentes y relaciones, se presentó además un ejemplo de instanciación. Tal instanciación se denominó

(10)

Doctus, el cual consiste en una herramienta de autor para cursos adaptativos. Doctus no solamente sirvió para ejemplificar el uso del modelo de referencia en su totalidad, sino también para refinar los sub modelos y algunos procedimientos involucrados. Como parte final de esta tesis, se realizó también la implementación y validación preliminar de Doctus. Esto se hizo con 51 sujetos, todos profesores en diversos niveles de formación. Los resultados obtenidos en esta etapa fueron sobresalientes en el sentido que todas las funcionalidades adaptativas fueron bien evaluadas y todos los encuestados manifestaron su entusiasmo por contar con una herramienta que les ayudara en sus prácticas docentes considerando a sus estudiantes como individuos particulares.

(11)

Modelo de referencia para sistemas educacionais adaptativos inteligentes

suportados por objetos de aprendizagem

RESUMO

A aprendizagem assistida por computador, conhecida mais amplamente com o nome genérico de e-learning, converteu-se numa poderosa ferramenta com amplas potencialidades dentro do campo educativo. Mesmo assim, uma das maiores críticas que esta recebe é que na maioria dos casos os cursos que são implementados seguem um enfoque “one size fits all”, isto é, que todos os alunos recebem exatamente o mesmo conteúdo e da mesma maneira desconhecendo suas necessidades particulares. Esta falha radica não só na falta de interação direita entre aluno e tutor, senão também na falta de um desenho instrucional apropriado que considere alguns dos diversos enfoques disponíveis hoje em dia.

Existem diversos enfoques que procuram solucionar este problema e adaptar o processo de ensino os estudantes. Pode-se dizer que na vanguarda de estes enfoques encontram-se os Sistemas Educacionais Inteligentes Adaptativos, os quais combinam as funcionalidades de dois enfoques: os Sistemas Hipermídia Educacionais Adaptativos y os Sistemas Tutoriais Inteligentes. Embora, logo de uma extensa revisão bibliográfica, se encontrou que existe ainda um inconveniente importante com este tipo de sistemas e em particular com seus modelos de referência: ou são demasiado simples, incluindo somente umas poucas funcionalidades; ou são demasiado complexos, o que dificulta seu desenho e implementação. Considerando este panorama, o objetivo principal de esta tese foi a definição de um modelo de referência intentando alcançar esse equilíbrio esquivo, de tal maneira que permita o desenho de cursos que se adaptem de una maneira efetiva e inteligente ao progresso e características de cada estudante, mas sem ser demasiado complexo. Outra propriedade importante desse modelo és que integra o uso de Objetos de Aprendizagem, promovendo assim a flexibilidade e a usabilidade.

Para alcançar este objetivo geral, três sub modelos foram considerados: um modelo do domínio, um modelo do estudante y um modelo do tutor. O primeiro serve para estruturar o domínio de conhecimento e foi definido usando a noção de objetivo de aprendizagem junto com um esquema flexível multi-nível com operações opcionais de pré-requisitos. O segundo visa caracterizar aos estudantes e considera informação pessoal, de conhecimento e psico-cognitiva. O terceiro pode ser considerado como o coração do sistema e define as funcionalidades adaptativas consideradas: sequenciamento y navegação, apresentação de conteúdo, evacuação, y suporte colaborativo.

Com o fim de clarificar os três sub modelos, assim como todos seus componentes e relações, se presentou um exemplo de instanciação que se denominou Doctus, o qual

(12)

consiste em una ferramenta de autor para cursos adaptativos. Doctus não somente serviu para exemplificar o uso do modelo de referência em sua totalidade, mas também para refinar os sub modelos e alguns procedimentos involucrados. Como parte final de esta tese, se realizou também a implementação e validação preliminar de Doctus. Isto foi feito com 51 sujeitos, professores em diversos níveis de formação. Os resultados obtidos em esta etapa foram sobressalientes no sentido que todas as funcionalidades adaptativas foram bem avaliadas e todos os pesquisados manifestaram seu entusiasmo por contar com uma ferramenta que lhes ajudara em seus práticas docentes considerando a seus estudantes como indivíduos particulares.

(13)

CONTENTS

ACRONYMS ... 1 LIST OF FIGURES ... 2 LIST OF TABLES ... 4 1 INTRODUCTION ... 5 1.1 Contextualization ... 5 1.2 Conceptual framework ... 6

1.2.1 Learning Management Systems ... 7

1.2.2 Adaptive Learning Systems ... 8

1.2.3 Intelligent Computer Aided Instruction ... 11

1.2.4 Adaptive and Intelligent Educational Systems ... 12

1.2.5 Learning Objects... 14

1.3 Research problem ... 15

1.4 State of the art ... 17

1.5 Research hypothesis ... 23 1.6 Thesis objectives ... 23 1.7 Scope ... 24 1.8 Stages ... 25 1.9 Contributions ... 26 1.10 Document’s outline ... 28 2 DOMAIN MODEL ... 31

2.1 Domain Model schema ... 31

2.2 Prerequisites structure ... 33

2.3 Domain Model instantiation ... 35

2.4 Chapter reflection ... 37 3 STUDENT MODEL ... 39 3.1 Personal information ... 39 3.2 Knowledge information ... 40 3.3 Psycho-Cognitive information ... 41 3.4 Other information ... 43

3.5 Student Model instantiation ... 44

3.6 Chapter reflection ... 48

4 TUTOR MODEL ... 49

4.1 Adaptive sequencing and navigation support ... 49

4.1.1 Learning Goals’ specification ... 50

(14)

4.2 Adaptive presentation ... 53

4.2.1 Incorporation of Learning Objects ... 53

4.2.2 Selection of Learning Objects ... 55

4.3 Adaptive assessment and feedback ... 58

4.3.1 Computerized Adaptive Testing and Item Response Theory ... 59

4.3.2 Assessment process ... 61

4.3.3 Feedback process ... 65

4.4 Adaptive collaboration support ... 65

4.4.1 Colleague searching for learning assistance ... 66

4.4.2 Group composition for collaborative learning activities ... 68

4.5 Tutor Model instantiation ... 72

4.5.1 Personal characteristics used for adaptive presentation ... 73

4.5.2 Psycho-cognitive characteristics used for adaptive presentation ... 73

4.5.3 Metadata standard for the Learning Objects... 74

4.5.4 Psycho-cognitive characteristics used for the colleague searching ... 75

4.5.5 Characteristics used for the group formation ... 76

4.5.6 Heuristic search method for the group formation procedure ... 76

4.6 Chapter reflection ... 77

5 IMPLEMENTATION AND VALIDATION ... 79

5.1 Hardware and software architecture ... 79

5.2 Application features ... 82 5.3 Validation ... 88 5.4 Chapter reflection ... 95 6 CONCLUDING REMARKS ... 97 6.1 Conclusions ... 97 6.2 Future work ... 99 6.3 Scientific divulgation ... 100 REFERENCES ... 103 APPENDIX A ... 113 APPENDIX B ... 121 APPENDIX C ... 125

(15)

1

ACRONYMS

3PL Three Parameters Logistic

AEHS Adaptive Educational Hypermedia Systems AHS Adaptive Hypermedia Systems

AI Artificial Intelligence

AIES Adaptive and Intelligent Educational Systems

AIWBES Adaptive and Intelligent Web-Based Educational Systems ALS Adaptive Learning Systems

AWBEHS Adaptive Web-Based Educational Hypermedia Systems CAD Computer Aided Design

CAL Computer Aided Learning CAT Computerized Adaptive Testing

DM Domain Model

HS Hypermedia Systems

HTML HyperText Markup Language

ICT Information and Communication Technologies ICAI Intelligent Computer Aided Instruction

ICC Item Characteristic Curve IRT Item Response Theory ITS Intelligent Tutoring Systems JSP Java Server Page

LMS Learning Management Systems

LG Learning Goal

LO Learning Object

LOM Learning Object Metadata SDG Simple Directed Graphs

SM Student Model

TM Tutor Model

UML Unified Modeling Language XML eXtensible Markup Language

(16)

2

LIST OF FIGURES

Figure 1.1: Taxonomy of CAL approaches ... 7

Figure 1.2: AHS general schema ... 9

Figure 1.3: ITS general schema ... 12

Figure 1.4: AIES typical functionalities ... 13

Figure 1.5: AIES dilemma ... 16

Figure 1.6: AHAM general scheme (Wu, 2002) ... 18

Figure 1.7: LAOS layers (Cristea & Mooij, 2003) ... 20

Figure 1.8: General schema of the proposed model ... 28

Figure 2.1: DM general schema ... 32

Figure 2.2: Typical DM schemas ... 32

Figure 2.3: Example of DM prerequisite definition ... 34

Figure 2.4: Example of DM ... 35

Figure 2.5: Example of DM using directory structure in Doctus ... 36

Figure 2.6: Examples of a book-like arrangement for DM ... 36

Figure 2.7: Example of DM with prerequisites ... 37

Figure 3.1: Overlay, differential and perturbation models ... 40

Figure 4.1: Example of DM prerequisite definition ... 52

Figure 4.2: Student – LO matching example 1 ... 54

Figure 4.3: Graphical representation of the psycho-cognitive characteristics... 56

Figure 4.4: Student – LO matching example 2 ... 57

Figure 4.5: Graphical representation of student and LOs from example ... 58

Figure 4.6: 3PL typical ICC ... 61

Figure 4.7: 3PL ICC varying parameters b (left) and a (right) (Baker, 2001)... 61

Figure 4.8: LG composition example ... 64

Figure 4.9: First stage in searching for learning assistance ... 67

Figure 4.10: Curry’s Onion model of learning styles (Curry, 1983) ... 73

Figure 4.11: General genetic algorithm schema ... 77

Figure 5.1: Doctus architecture ... 80

Figure 5.2: Entity-relationship model of Doctus database ... 82

Figure 5.3: Learning goal composition and prerequisites structure in Doctus ... 82

Figure 5.4: Structure of the example course ... 83

Figure 5.5: Activities definition in Doctus ... 83

Figure 5.6: Assessment item bank creation in Doctus ... 84

Figure 5.7: Course presentation in Doctus ... 85

(17)

3

Figure 5.9: Collaborative activity in Doctus ... 86

Figure 5.10: Assessment process in Doctus ... 88

Figure 5.11: Validation session ... 89

(18)

4

LIST OF TABLES

Table 1.1: LMS list ... 8

Table 1.2: Traditional CAL versus STI ... 11

Table 1.3: LO metadata initiatives ... 14

Table 1.4: LO repositories ... 15

Table 2.1: Example of LG description table in Doctus ... 35

Table 2.2: Example of LG connections table in Doctus ... 35

Table 2.3: Example of LG pre-requisites table in Doctus ... 37

Table 3.1: General knowledge information in the SM ... 41

Table 3.2: List of learning styles models in implemented systems ... 42

Table 3.3: Other feasible information in the student model ... 44

Table 3.4: Personal information in Doctus ... 44

Table 4.1: General specification for a LG ... 51

Table 4.2: LG definition example ... 51

Table 4.3: Guessing probabilities with regard to the type of question ... 62

Table 4.4: Example data with four students and two attributes... 71

Table 4.5: Example of scaled values ... 71

Table 4.6: Example of feasible solutions... 71

Table 4.7: IEEE LOM categories ... 74

Table 5.1: Usability test questionnaire ... 90

(19)

5

1

INTRODUCTION

This chapter, as its name implies, introduces this thesis presenting the motivation, conceptual framework, problem description, and state of the art. Later it defines thesis scope, this is, the research hypothesis, objectives and contributions.

1.1

Contextualization

It is not a secret that XXI century is the era of the knowledge and information-based economy, and that society progress is more dependent of science and technology development than in any other moment in history, being the intellectual resources the major source to promote innovation. This insatiable need for knowledge represents important challenges in the process of education, training, updating an improvement of skills, not just for the academic field, but also for industry and society in general. As it is mentioned in (LearnFrame pp.17-18, 2000):

Where the resources of the physically-based economy were coal, oil, and steel, the resources of the new, knowledge-based economy are brainpower and the ability to effectively acquire, deliver and process information. Those who are effectively educated and trained will be the ones who will be able to economically survive and thrive in our global, knowledge-based economy. Those who don't will be rendered economically obsolete.

An alternative to solve this increasing need of knowledge acquisition is the Computer Aided Learning (CAL) which has become very popular in the last years thanks to its principle of flexible access anytime and anywhere. Among CAL’s main strengths one could mention: a) it increases availability of learning experiences for those students who cannot or chose not to assist to traditional face-to-face classrooms; b) it allows for the development and divulgation of instructional content in an efficient way in terms of cost and; c) it allows for increasing the coverage of students without a deterioration in the education quality.

To strength those statements it is important to mention that in United States, approximately 3.9 millions of people studied in 2007 university on-line courses, 12% more than previous year, whereas the whole university population grew 1.2% according to data

(20)

6

from Sloan Consortium1. In this country, the National Center for Education Statistics estimated that number of public school students who enlisted in technology-based distance courses grew around 65% between 2002-03 and 2004-05. In a more recent study presented in (Picciano & Seaman, 2009) it was estimated that more than a million of K-12 students took online courses during 2008 and 2009.

The main question that these data lead to, is if this approach is more effective than traditional, face-to-face education, and if it is not, why this tendency has appeared. According to a study that the SRI International2 consultant made for the United States Department of Education, technology-based education is in fact more effective, with a small difference in favor when it is completely virtual, but quite bigger when it refers to projects that combine traditional classes with virtual formation using new technologies. It is not, as conclusions of such study say, that computers have some sort of magical effect, or that model itself is more effective. Instead it states that the use of such tools in education usually implies that student spends more time studying, looking for additional information for his/her own, sharing it, and collaborating with classmates. In summary, being more prone to take the lead of his/her own learning instead of being a passive individual most of the time anonymous in a crowded classroom (El País, 2009).

In tune with this affirmations, in the survey presented in (U.S. Department of Education, 2009) a systematic analysis was made about the researches in this topic between 1996 and 2008. Such a survey selected 99 studies which made a reliable quantitative comparison among the two kinds of teaching, choosing finally 49, most of them very recent. Assigning them values to the learning difference, measured throughout reliable test, the central outcome was that entirely CAL produced a slightly better effect than traditional teaching.

1.2

Conceptual framework

CAL refers to the use of computers as a key element within educational environment. Although this definition may cover the general use of computers within a traditional classroom, it is more accepted that it refers specifically to a structured environment where computers are used explicitly for the teaching process, being the students an active part on it. Another very popular term associated to CAL is e-learning, which, even if does not have a universally accepted definition, is usually related with distance education supported by Information and Communication Technologies (ICT).

Going back in history, it is important to mention that first CAL systems dated from early 50’s, more known as Computer Aided Instruction or Computer Based Training, were characterized for being more focused into just instruction than in actual teaching. The functionality of these early systems was very restricted to the software and hardware of that time: interaction with user was made through terminals and there was very poor processing

1

www.sloan-c.org 2

(21)

7 and storage capacity. One of the more known examples is PLATO: Programmed Logic for Automated Teaching Operations, which was developed in the Illinois University with the aim of teaching courses in a massive and automatic way.

Until 80´s, most of these systems characterized for teaching in a very procedural way, with no personalization features, and in an unfriendly manner. From then, CAL has evolved a lot, promoted by ICT reception not only to complement traditional classrooms but, as mentioned before, to reach more students (probably located geographically far away) and in a better way. Such evolution allowed for the emergence of several approaches with their own particularities, being their differences unknown in many cases for teachers and instructional designers.

Being aware of those differences, a taxonomy of several of these approaches is presented in this section, describing each one in a brief but concise way. For the sake of a better understanding it was divided in four major trends as it is presented in figure 1.1, presenting their most standing features and, where it has been possible, listing some studies and implemented systems.

Figure 1.1: Taxonomy of CAL approaches

1.2.1 Learning Management Systems

The Learning Management Systems (LMS), also known as Course Management Systems, are web based platforms whose main features are to manage, monitor and report interaction of students with the learning material, with teacher and with other students. In order to do that, most LMS generally use a client-server architecture where teachers configures applicative interface using web forms to make course contents available. This architecture and mode of use has allowed the overcrowding of LMS, promoting the rising of many robust commercial implementations.

CAL LMS ICAI ITS ALS AEHS AWBEHS AHS AIES AIWBES LO

(22)

8

The LMS also characterize for providing a large set of tools to assist the development of courses. Among these tools one could mention files manager, forums, chat, calendar, automatic assessment questionnaires, and statistics of use, among others. All of them are precisely what make that LMS, although they were originally designed to develop on-line courses, are being used for many institutions to complement face-to-face classrooms, facilitating the teachers’ labor, centralizing resources and serving as meeting point for students.

Table 1.1 shows a list of some of the most popular LMS. A detailed comparison of some of them, including functionalities and technical specifications may be found in (WebCT, 2008).

Table 1.1: LMS list

Name URL License type

Amadeus amadeus.cin.ufpe.br Free

Angel learning www.angellearning.com Integrated to Blackboard BlackBoard www.blackboard.com Proprietary

Claroline www.claroline.net Free

Dokeos www.dokeos.com Free

ILIAS www.ilias.de Free

Joomla www.joomlalms.com Proprietary

Moodle www.moodle.org Free

OLAT www.olat.org Free

Sakai www.sakaiproject.org Free Schoolar360 www.scholar360.com Proprietary Sharepoint www.sharepointlms.com Proprietary

WebCT www.webct.com Integrated to Blackboard

1.2.2 Adaptive Learning Systems

The Adaptive Learning Systems (ALS), also known as Adaptive Learning Environments or Adaptive Courseware Environments, refer in general to those systems that presents a knowledge domain to students in an adapted way, under the principle that it increases significantly learning speed (Davidovic et al., 2003). In these systems the adaptation scope is manly related to preferences and characteristics of students. Preferences are related to student’s likes in his/her role of a computer system user: colors, sizes, fonts, etc; whereas characteristics are related to educational processes: knowledge level, learning goals, etc. To contrast all this information with the knowledge domain, two adaptation levels are usually considered: content and links. The first one is known as adaptive presentation and the second one as adaptive navigation support.

Within the ALS, the Adaptive Educational Hypermedia Systems (AEHS) as its name implies, are a particular approach whose presentation structure is based on hypermedia content (hypertext + multimedia). As it is shown in figure 1.1, they are directly related to the Adaptive Hypermedia Systems (AHS) which have been widely used as presentation tools for personalized content. The general schema for these systems, presented in figure

(23)

9 1.2, is the same as the AHS because AEHS may be considered as a specific application with the difference that students are the users of the system so models are related to their learning process.

In this point, it is important to clarify the difference between two terms: adaptable and

adaptive. In one hand, systems that allow the user to change certain parameters and adapt their behavior accordingly are called adaptable. In the other hand, systems that adapt themselves to the user automatically, based on the assumptions they make about user needs are called adaptive (Opperman et al., 1997). Considering this difference, when using the term adaptation in the rest of this document, it will refer to the second term.

Figure 1.2: AHS general schema

Knutov et al. (2009) propose six questions to explain general purpose of adaptation in AHS, from which, the next ones were formulated:

 From what adaptation could be made? (from what?)

 To what adaptation could be made? (to what?)

 Why adaptation is required? (why?)

 What could be adapted? (what?)

 When adaptation could be used? (when?)

 How adaptation could be made? (how?)

 Where adaptation could be used? (where?)

Whose answers are related to models presented in figure 1.2. The final application is based on the Domain Model (DM) that describes how conceptual representation of the domain is structured. In other words, DM usually answers question “from what?”, indicating the elements that composes such domain as well as their relationships.

The User Model usually answer question “to what?” giving information about user preferences and characteristics. This model may also help to answer question “why?” providing information about user goals.

Final application must adapt instruction, content, presentation and navigation to user and, in order to do that, the Adaptation Model must communicate with the other two to answer questions “what?” along with “when?” and “how?”.

Domain Model User Model

(24)

10

Finally, “where?” may be understood as a more general question that refers to the AHS application area, this because it does not have to be necessarily educational. Other potential applications are information systems, personalized views, help systems, etc.

AEHS has as advantage that it allows developing non-linear interactive applications and admits a direct link between adaptation techniques and user interface. A classification of these techniques for the two previously mentioned adaptation levels may be found in (Brusilovsky, 2001):

Adaptive presentation

o Text presentation: extension, detail level, contextual information, etc. o Multimedia objects presentation: format, size, quality, etc.

o Mode: selection of one or more objects according to user features Adaptive navigation support

o Direct guidance: insertion of “next” type links

o Ordering: links localization according to some criterion (relevance for instance)

o Hiding: restricted access to certain contents

o Formatting: changing in links appearance to denote some special feature like visited, non-visited, recommended, optional, etc.

o Generation: insertion of extra links

o Navigation map: graphical representation of hyperspace

Some widely documented AEHS are: ESCA (Grandbastien, & Gavignet, 1994), SYPROS (Gonschorek & Herzog, 1995), ELM-ART (Brusilovsky et al., 1996), Hypadapter (Hohl et al., 1996), Hypercase (Micarelli & Sciarrone, 1996), InterBook (Brusilovsky et al., 1998) and KBS-Hyperbook (Henze & Nejdl, 2001).

As it may be seen in figure 1.1, AEHS have a subdivision known as Adaptive Web-Based Educational Hypermedia Systems (AWBEHS) which are focused specifically on the web so users (students) access them throughout a web navigator. This approach is in fact quite natural for this kind of applications considering that web is based in languages like HTML and XML that facilitates some fundamental tasks from AEHS about links schema. Another advantage is that it allows accessing in real time to applications from any equipment in a local network or over Internet.

One disadvantage of this approach however, compared to non-web AEHS, is that these last ones may have a more strengthen relationship between interface and underlying functionality, this is, every user action may be recorded: every mouse movement, scrolling, window size change, etc., and such information may be used for adaptation purposes (De Bra et al., 2004).

(25)

11 Some well-known implemented AWBEHS are: AHA (De Bra & Calvi, 1998), AHM (Da Silva et al., 1998), TANGOW (Carro et al., 1999), ECSAIWeb (Sanrach & Grandbastien, 2000), SmexWeb (Albrecht et al., 2000), AIMS (Aroyo & Dicheva, 2001), NetCoach/ART-WEB (Weber et al., 2001), AHA! 2.0 (De Bra, et al., 2002), MetaLinks (Murray, 2003), CoMoLe (Martín et al., 2006) and GOMAWE (Balik & Jelinek, 2007).

1.2.3 Intelligent Computer Aided Instruction

The Intelligent Computer Aided Instruction (ICAI), also known as Intelligent Computer Aided Learning or Intelligent Learning Environments emerges as a natural evolution of first CAL systems providing an individualized learning experience for student simulating interactions with a real teacher. Within this context, when talking about individualized or personalized instruction it is understood that the system does not treat all students equally, so they do not receive the same content in the same time nor in the same way. To achieve such a task, these systems represent in a separate way the content, the teaching strategies and the student characteristics.

Within ICAI, one of the most known approaches are the Intelligent Tutoring Systems (ITS). Even if their more general definition is that they are tutoring systems that have incorporated intelligent components, commonly associated to Artificial Intelligence (AI) techniques, some authors extends such definition adding that they may count with procedures and representations of knowledge from the computational linguistics and cognitive sciences fields (Samuelis, 2007). ITS are a well-known approach from ICAI and may be described as computer systems that try to imitate a human tutor generating interactions when they are required by students, as well as detecting individual learning problems and providing means to solve them. In this sense this kind of systems are quite different from early CAL systems (Vicari & Giraffa, 2003). Some of these differences are presented in table 1.2.

Table 1.2: Traditional CAL versus STI

Traditional CAL ITS

Theorical basis Skinner theory (behaviourism)

Cognitive psychology System schema One only structure defined in

an algorithmic way

Structure subdivided in models

Content sequencing Fixed Heuristic

Student modeling Validation of final answers Validation of the whole student – system interaction

Instruction features Tutorial, exercises Socratic, interactive environment, guidance

Although there is not an explicit consensus about ITS components, most authors agree that they have a general schema like the one presented in figure 1.3, which is consistent with the previously mentioned CAL separation principle.

(26)

12

Figure 1.3: ITS general schema

Knowledge is structured in the Domain Model (DM) and its representation would depend manly of the general kind of such knowledge, this is, if it is factual, relational, procedural, analytic, etc. Some common representation forms are directed graphs, hierarchical trees, semantic networks, production rules, expert systems, etc.

In the Student Model (SM) is where all individual student characteristics are stored. As it is mentioned in (Eyharabide et al., 2009) information in this model may be divided in several categories like personal (name, age, gender, etc.), cultural (race, residence region, etc.), technological (access device, bandwidth speed, etc.) and system interaction (accessed content, number of sessions, etc.). Several works add to this information some other categories like environmental conditions, emotional, personality-related, among others.

The Tutor Model (TM), also known as Pedagogical Model, is the one in charge of guiding the teaching process, deciding which pedagogical actions must be performed, as well as how and when that must be done. In other words this model deals with delivering the didactical strategies that are specified in the system in an adapted way to the student needs (based on SM), considering the knowledge domain (from DM).

Finally, the Interface Model determines how activities and contents are presented in the screen to each student. This model deals with the lower level interaction details like files formats, links, buttons, forms, etc.

A very extensive list of implemented ITS and tools to develop them, as well as the corresponding references, is found in (Murray, 1999).

1.2.4 Adaptive and Intelligent Educational Systems

Figure 1.1 shows that Adaptive and Intelligent Educational Systems (AIES) may be considered as an intersection between AEHS and ITS. More specifically, as it is show in figure 1.4, it may be said that from AEHS they inherit adaptive presentation and adaptive navigation support, described in section 1.2.5, whereas from ITS they usually incorporates some of the next functionalities (Brusilovsky & Peylo, 2003; Peña, 2004):

Domain Model Student Model

Tutor Model

(27)

13

 Curriculum sequencing: suggesting to student the “optimal” learning path, understood as the planed sequence of activities and contents that he/she must accomplish within the knowledge domain.

 Adaptive assessment: in the same way that instruction should be provided in an individual basis, assessment should be too. The most common way to do that is presenting assessment items in a sequence that is dependent on the correctness of the examinee’s responses, looking for an accurate measure of his/her achievement level.

 Intelligent analysis of solutions: more than assessment, its goal is to discover the mistakes committed by student, e.g. misconception, miscalculation, etc., looking for plausible causes with the aim of helping him/her to correct them.

 Problems solving support: to provide intelligent help, e.g. giving advices, reminders, etc., to student when he/she faces a specific activity. This functionality differs from the previous one because it is not remedial, so is not performed just when a mistake in the student reasoning is detected, but as some sort of continuous guidance.

 Adaptive collaboration support: to use the system’s knowledge about students to facilitate collaborative learning activities. Examples include forming a group for collaborative problem solving at a proper time, or finding the most adequate peer to answer a doubt about a specific topic.

Figure 1.4: AIES typical functionalities

In the same way that in AEHS, the AIES have a subdivision called Adaptive and Intelligent Web-Based Educational Systems (AIWBES) whose functionalities and technical issues, as its name implies, are related specifically with web format. As mentioned in (Keleş et al., 2009), this is the more common trend, and there are several successful efforts to translate existent systems to the web world (Ritter, 1997; Alpert et al., 1999), whereas there are many other whose since their conception were designed to run in this environment (Chen, 2008; Lin et al., 2008).

Some well-documented AIWBES are: AdaptWeb (Oliveira et al., 2003), MAS-PLANG (Peña, 2004) and ZOSMAT (Keleş et al., 2009). In the regional scope one could mention:

Adaptive presentation Adaptive navigation support Curriculum sequencing Intelligent analysis of solutions Problems solving support ITS AEHS AIES Adaptive collaboration support Adaptive assessment

(28)

14

SICAD (Duque, 2005), ALLEGRO (Jimémez, 2006), AMPLIA (Vicari et al., 2008) and CIA (Moreno et al., 2009), which have some of the functionalities described previously.

1.2.5 Learning Objects

Learning Objects (LOs) are presented in figure 1.1 as a separated “bubble” in the CAL context for a simple reason: More than being an approach to create learning systems, they may be considered as an alternative to represent, and finally to store educational content.

Although there are a lot of definitions of what a LO may be, in a very concise way it can be said that is any digital resource that is used in a simple or composite way to support teaching/learning process and that may be re used. A very common metaphor that is used to explain LOs and to extend previous definition is the LEGOs blocks: little instructional pieces (LEGOs) that may be assembled between them in a bigger instructional structure (a castle for example) and that may be reused later in other structures (a spaceship for example).

This analogy, although is very illustrative, has the next conceptual problems related to LOs properties (Wiley, 2001): a) any LEGO block may be combined with any other; b) LEGO blocks may be assembled in any way; c) LEGO blocks are very simple so even a child may combine them. Considering these issues, the same author proposes the atom as a new metaphor for LOs: an atom is something little that may be combined and recombined with others to form bigger structures (molecules). This metaphor is more harmonious with LOs properties: a) not any atom may be combined with any other; b) atoms only may be assembled among them depending of their own internal structure; c) some training is required to combine atoms. Summarizing, these properties mean that structuring of educational content from LOs is possible as long as there is an appropriate instructional design in the middle.

One important feature of LOs is that they may be described through metadata that facilitate their administration. Such metadata may be defined by standards, being the more known Learning Object Metadata (LOM) from IEEE, although there are also several initiatives known as specifications, which procure to capture a consensus between researchers summarizing or extending certain aspects of an existent standard. A list of some of those initiatives for specific regions, countries or research centers and universities is presented in table 1.3.

Table 1.3: LO metadata initiatives

Name Comunity URL

Dublin Core Internacional http://dublincore.org/ UK LOM Core United Kingdom http://metadata.cetis.ac.uk

CanCore Canada http://cancore.athabascau.ca

ANZ-LOM Australia and New Zealand http://www.thelearningfederation.edu.au

OBAA Brazil http://www.portalobaa.org

NORLOM Norway http://www.itu.no/no/NSSL

(29)

15

Considering their features, and particularly their reusability philosophy, lots research groups and institutions have adopted LOs as an importer part of their CAL processes and, as a consequence of that, several sources have emerged and became of huge help for the educational community. Some of those sources, known as repositories and federations are listed in table 1.4.

Table 1.4: LO repositories

Repository URL

ARIADNE http://www.ariadne-eu.org

ALI: Apple Learning Interchange http://ali.apple.com CAREO: Campus Alberta Repository of

Educational Objects

http://www.ucalgary.ca/commons/careo FEB: Federação de Repositórios Educa

Brasil

http://feb.ufrgs.br LA FLOR: Latin American Federation of

Learning Object Repositories

http://laflor.laclo.org LORN: Learning Object Repositories

Network

http://lorn.flexiblelearning.net.au MELOR: Medical Learning Object

Repository

http://gilt.isep.ipp.pt:8080/melor MERLOT: Multimedia Educational

Resource for Learning and Online Teaching

http://www.merlot.org

MIT Open Courseware http://ocw.mit.edu

Wisc-online http://www.wisc-online.com

1.3

Research problem

After the conceptual framework analysis presented in section 1.2, the next reflections were extracted:

1. Even if the LMS are the kind of applications more used by educational institutions of different formation levels around the world thanks to their robustness and ease of use, they are support platforms which do not provide an actual individualized teaching, or at least not in their original commercial versions.

2. Differently to LMS, ALS and ICAI are approaches that have as main goal to provide an individualized learning experience, being in this way a lot more appealing. Both approaches however have as disadvantage that they usually do not count with generic authoring tools of ease use, due to their conceptual and functional complexity. According to Moundridou & Virvou:

(30)

16

The main flaw of ITSs and possibly the reason for their limited use in workplaces and classrooms is the complex and time-consuming task of their construction (2003, p. 158).

Consequently with this affirmation Woolf & Cunningham (1987) estimate that an hour of educational material for an ITS for example requires more than 200 hours of development time. Although such estimation may have been reduced due to the technology advances, the truth is that proportion is still high.

3. Although it could be said that ICAI systems are more advanced than ALS from the pedagogic point of view, it is clear that there are some conceptual similarities among them. As it can be seen in figures 1.2 and 1.3 both approaches are based in an architecture that distinguishes two models: a Domain Model – DM and a Student Model – SM. Both also have a third model. In the case of ICAI, and particularly in ITS, it refers to the Tutor Model – TM, whereas in ALS to the Adaptation Model. Even if they are not the same because in the case ITS it covers a larger spectrum of processes, in some AEHS the authors particularized this model giving it similar titles as Teaching Model (De Bra et al., 1999), Pedagogic Model (Henze, 2000), or Narrative Model (Conlan et al., 2002).

4. Precisely, due to the similarities described in previous reflection, and as a natural evolution of both approaches, it may be said that AIES are found in the vanguard because it combines some of their best functionalities. Even if this sounds very promising, it also implies that the complexity in its design and implementation is higher. In fact, if one wants to see a graphical representation of the relationship between these two dependent issues, it would be something like the one presented on figure 1.5.

Figure 1.5: AIES dilemma

C

ompl

exit

y

Adaptation and intelligent tutoring features Robustness

Simplicity of design and implementation

(31)

17 5. Independently of the approach, LOs seem to be a very useful alternative in CAL, particularly because their granularity, as well as the way they are developed and specified, promotes reusability. In Wiley’s words:

If learning objects ever live up to their press and provide the foundation for an adaptive, generative, scalable learning architecture, teaching and learning as we know them are certain to be revolutionized. (2002, p. 15)

From these reflections, it could be said that complexity in the design and implementation of the educational systems in the analyzed approaches increases as they have more adaptation and intelligent teaching features with the aim of providing a more individualized learning experience. Being so, the research problem may be defined as providing mechanisms that allow the development of this kind of systems but achieving equilibrium between the two issues mentioned. In other words, the problem is how to define a reference model for AIES, without having too much complexity, but detailed enough to facilitate subsequent implementations which would represent powerful tools in educational context.

It would be even more helpful if such model contemplates LOs as part of its foundations. Besides the advantages described in section 1.2.11, there are lots of reasons for choosing LOs to encapsulate educational content, being three of them particularly interesting. The first one is that they allow separating the knowledge domain structure from actual content, providing flexibility and, as mentioned before, reusability. The second one is that metadata that describe them may be used not only for characterizing purposes but also for adaptation. And third is that there are nowadays a considerable number of LOs repositories, many of them with open use licenses, where teachers and designers may access thousands of them.

1.4

State of the art

Considering that in research literature around AIES there is a very significant number of studies and publications, the next selection criteria were chosen for the state of the art: a) they may come from any of the approaches described in section 1.2, always as they cover explicitly the user, i.e. student, adaptation; b) they should present a clear separation of the systems components that are considered to achieve such adaptation; and c) they should include an adequate description with a considerable detail level about those components and their relationships. This criterion excludes platforms, systems and specific tools from which their design is unknown or non-properly described.

Within these criteria it does not appear the explicit use of LOs even if they are a fundamental part of this research. Such situation is due to LOs are a relatively new concept and very few studies use them considering the previous criteria. This however is not necessarily an obstacle because, as it will be seen later, many issues about adaptation may be extrapolated to consider them.

(32)

18

The studies that accomplish the mentioned three criteria are described here in chronological order.

AHAM (Adaptive Hypermedia Application Model) (Wu, 2002) is a reference model for the design of AHS which, even if are not focused specifically in the educational context, servers as an important reference point because it aims to provide personalized views to the user based on his profile. In other words this model may be useful for the AEHS design but from a higher abstraction level considering that its main goal is not learning but just information transmission. This study is an extension of the Dexter model (Halasz & Schwartz, 1990) which is in turn a well formalized reference model but for conventional Hypermedia Systems (HS). It adopts the AHS definition provided by (Brusilovsky, 1996) which states that it is any HS that captures some user characteristics and use them to adapt several system visible features.

AHAM defines three sub models that coincides with the ones presented in figure 1.2, which together conform what in Dexter model is called the storage level. The final goal of these models, as author expresses, is to describe the structure and functionality of the designed AHS as well as to make the communication, or more precisely the translation, between them possible. As it is shown in figure 1.5, in the lower part of the complete scheme, the content is found (in the within-component layer), whereas its relationship with the storage layer is made throughout connection points defined in the anchoring layer. The ‘T’ structure in the storage level is due to the boundaries and interaction among the three sub-models. Finally, in the upper part of the scheme, the run-time layer is found where the final user presentation is located.

Figure 1.6: AHAM general scheme (Wu, 2002)

Similar to AHAM, Munich (Koch & Wirsing, 2002) is a reference model for AHS and is also considered an extension of the Dexter model to which incorporates some user modeling issues and rules-based adaptation mechanisms. The main feature of this study is that proposes an object oriented specification described in UML which integrates an

(33)

19 intuitive visual representation with a formal OCL (Object Constraint Language) description.

The Goldsmith model (Ohene-Djan & Fernandes, 2002) is similar to the previous two although is not based in the Dexter model and even presents a comparison between them arguing that its scope is wider. It consists in the description of an abstract model which defines the functionality of a core for HS as well as the specification that allows its personalization. Such a model is composed of functions which are divided in three types or regions: the ones from the H-region that models the non-customizable user-application interaction where a formal specification of the hyper-pages that stores information is required; the ones of the P-region that models content personalization performed in an explicit way for the user through an annotation and rewriting processes which are later translated to an also formal language; at last, the ones of the A-region that models the content adaptation which is performed in an autonomous way by the system.

In (Cristea & Aroyo, 2002) a model to design authoring tools for AWBEHS is presented. Such a model is composed of three layers, the first one is the conceptual layer that expresses the DM and is divided in two sub layers, one for atomic concepts, understood as a part of the knowledge domain, an another one for composed concepts. Second layer contains lessons that are analogous to a chapters and sub-chapters structure, and represent the order and manner in which concepts are presented. Third layer is also divided in two sub-layers, one for student adaptation that specifies which material must be presented under what set of conditions, and another for presentation that specifies the format in which information as such is showed in the web pages.

LAOS (Cristea & Mooij, 2003) is a generalization of the previous work because is not focused specifically in educational systems but has AHS in general. Besides the AHS general models, this research proposed two new ones: the Restriction and Goals Model and the Presentation Model. The first one tries to provide a presentation that is more focused in the instructional goals and at the same time limits the search space in the knowledge domain, whereas the second one takes in consideration the interface proprieties and provides a connection with the code generation for different platforms (HTML for instance). The general structure considering the five models is presented in figure 1.7. Besides this division, this research also proposes the use of operators to manipulate the elements in each model.

In (Cristea & Calvi, 2003) a study that somehow complements the previous one is presented. Although it does not consider de GM it does specify in detail the AM which divides in three levels. In the lower level the techniques for content presentation and navigation support are presented. In the middle level such techniques are grouped in typical adaptation mechanisms and for them, a set of rules and operators are build which defines a

(34)

20

programming language for the adaptation strategies. Finally, in the higher level the user’s cognitive styles are considered to determine the more appropriate didactical strategies.

Figure 1.7: LAOS layers (Cristea & Mooij, 2003)

GAM (Generic Adaptivity Model) (de Vrieze et al., 2004) starts from AHAM model but is more generic because it may be used also for non-hypermedia applications. Another two fundamental differences are that GAM has a lower abstraction level and proposes an additional model to describe application interface considering its connection with the AM. To specify the relationships among the four models, GAM is based on a states machine approach in which all user interaction with the application generates an event that may produce a change. Such a change in turn may be translated into the modification of a state.

In (Karampiperis & Sampson, 2005) a general AHS structure similar to AHAM is considered but focusing in the AM which subdivide in two processes: one for concept selection and another for content selection. The first one refers to the mapping of the learning goals with the concepts from the knowledge domain as the student advances trough course, whereas the second one refers to the resources that are selected to cover each concept based on the relationship between their educational features and the student’s cognitive characteristics and preferences.

(35)

21 To perform these processes, this study differentiates from others that consider predefined adaptation rules proposing a decision model which, based on the generation of all possible learning paths that maps certain learning goal, selects the more appropriate resources for each student.

In (Chen et al., 2006) a model for the curriculum sequencing in AIWBES is proposed using an Item Response Theory (IRT) approach. It considers several aspects as the difficulty level of the course, the student knowledge measured trough initial tests and the relationships between de domain concepts.

Although authors do not call it that way, in (Bouzeghoub et al., 2006) an ALS model supported in LO is proposed using a general schema with three models: the DM which in this case represents the concepts that are covered with the LOs; the SM where basically his/her preferences and progress are stored; and a Pedagogical Model which is in charge of presenting LOs to student. Another distinctive feature of this work is that it uses metadata for LOs and a RDF (Resource Description Framework) algebra for the operations that may be applied.

In (Curilem et al., 2007) a mathematical model, specifically under the finite automata approach, is described for the architecture and functionalities of ITS with the aim of facilitating the integration of its design between computing and pedagogy fields. This study considers two pedagogical strategies joint to three theories to build the tactics. Here, a strategy is understood as the set of conditions and stages that are needed for the teaching/learning process, whereas the tactics indicate how a strategy may be implemented.

TEx-Sys (Stankov et al., 2008) is a model for the construction of ITS based on pedagogical activities considering the next four phases cycle: didactic, perception, diagnosis and evaluation, and finally help and remediation. The didactic phase involves the specification of the knowledge domain, the student characterization and the adaptation methods for instruction according to the student needs. The perception phase deals with the student’s previous knowledge, whereas the diagnosis and evaluation phase deals with its evolution. In case of existing, the student’s conceptual errors activate the help and remediation mechanism with the aim of minimizing the difference between his/her knowledge and the taught domain.

Some interesting features of this study are that it uses semantic networks to formalize knowledge domain and has a simple procedural mechanism for adaptive assessment.

In (Chen, 2008) a model that is similar to (Chen et al., 2006), previously described, is presented, with the main difference that it does not use IRT for curriculum sequencing but proposes a genetic algorithm approach instead.

(36)

22

CIA (Moreno et al., 2009) may be considered within AIES approach because it considers content adaptation and adaptive assessment issues jointly with AI techniques. In this work the DM is modeled as a hierarchical tree with a specific structure where LOs are located in its leaves. Such LOs are described through a metadata standard and the adaptation process is based in one student’s cognitive characteristic, the learning styles, according to a specific model. For the implementation of the whole system, a software agents architecture is proposed where each one of them has a particular role: to represent systems actors (students and teachers), to manage the four main ITS models (domain, user, tutor, and interface), or to perform a specific sub process like adaptive assessment.

Similar to the previous study, SICAD+ (Duque, 2009) may be considered as an AIES focused mainly in the adaptation task and is supported by software agents. It differentiates however for being a lot more generic about domain structure, the pedagogical strategies that may be implemented and the student’s characteristics that are used to adapt content. The core of this study is a planner module that incorporates the adaptation strategy and translates the curriculum sequencing problem in a AI planning problem, which is solved using an algorithm called HTN (Hierarchical Task Network). Although in this study author does not talk about LO but educational content in general, it is clear that in this case they are analogous ideas and even author proposes the use of metadata standards that are used precisely in the LO world.

As a summary of this section, the next reflections may be highlighted:

The reference models for AHS and AEHS (Wu, 2002; Koch & Wirsing, 2002; Ohene-Djan & Fernandes, 2002; Cristea & Aroyo, 2002; Cristea & Mooij, 2003; Cristea & Calvi, 2003; Vrieze et al., 2004; Karampiperis & Sampson, 2005) have as advantage being very robust and formal about the components definition, functionalities and their relationships. Besides that, even when they are delimited to hypermedia applications, many of their techniques may be applied, or even more extrapolated, to other kinds of adaptive educational systems.

About the analyzed models for ITS (Curilem et al., 2007; Stankov et al., 2008) it may be said that, precisely for being focused in very complex systems, they are too general models which do not describe in a formal way all their functionalities. In contrast to this situation, works like (Chen et al, 2006; Chen, 2008) of the AIES approach, although they are quite detailed, are focused only in certain features of those systems, without specifying the relationship with the other components.

Other researches such of Bouzeghoub et al. (2006) that incorporate adaptation and LOs within the ALS approach are too simple, because consider only some adaptation issues, leaving apart some others that are equally important for all CAL systems like assessment for example.

(37)

23 More recent researches like the one presented in (Moreno et al., 2009) have the advantage of joining interesting issues as the consideration of student’s cognitive characteristics for educational content adaptation based on LOs. It has however as disadvantage that is not generic enough to help in the design of other applications and in which, neither the DM, nor the TM are flexible.

From all analyzed works, the one presented in (Duque, 2009) is perhaps the closer to what this thesis pretends, because it is a generic model for the creation of adaptive educational systems that uses LOs. There are however several fundamental differences that may be identified. The first one is that, even if it is a generic model, the detail level in which is described is closer to the analysis stage than to the design (from the software engineering point of view) and does not present a formalization of all involved tasks. The second one is that it uses a specific mechanism based on HTN for curriculum sequencing and content presentation, which may turn complex in the later implementation stage. The last one is that it is supported in a software agents architecture which, although it could be helpful, also increases the implementation complexity.

1.5

Research hypothesis

Enclosed within the research problem described in section 1.3 and considering the state of the art presented in section 1.4, the next research hypothesis is formulated:

It is possible to achieve a comprehensive design of personalized educational systems, which adapts to the student’s needs and characteristics, using AIES techniques as well as LOs to support the teaching/learning process.

1.6

Thesis objectives

In order to answer the research hypothesis, the next thesis’ general objective is formulated:

General objective

To define a robust reference model for the design of computer aided learning systems which adapt themselves to students and are supported by learning objects.

Figure

Figure 1.1: Taxonomy of CAL approaches
Table  1.1  shows  a  list  of  some  of  the  most  popular  LMS.  A  detailed  comparison  of  some  of  them,  including  functionalities  and  technical  specifications  may  be  found  in  (WebCT, 2008)
Figure 1.2: AHS general schema
Figure 1.3: ITS general schema
+7

References

Related documents

The present study aimed to evaluate the antibacterial effects of silver nanoparticles synthesized by Verbena officinalis leaf extract against.. Yersinia ruckeri, Vibrio cholerae

Quinazolines and its derivatives represent one of the most active classes of compounds, which possess wide range of biological activities like anti-bacterial,

in the back, and punk and garage rock bands in between, We would later play some of.. our most memorable shows at Stinky’s, but that night the band was Jackie

• The lack of labeled data is mitigated by using unsupervised ranking to approximate the true document space and design two deep generative models to leverage the contents of

The projected gains over the years 2000 to 2040 in life and active life expectancies, and expected years of dependency at age 65for males and females, for alternatives I, II, and

Demonstrate knowledge and understanding of the NSW Board of Studies Business Studies Syllabus for stage 5/6 and various Department of Education policies, particularly those

and subsidiaries as of December 31, 2008 and 2007, and the results of their operations, the changes in their stockholders' equity and the changes in their financial position for

fragrance-sensitive persons.ii In the absence of identifying the common allergens of fragrances, the fragrance-sensitive consumer is often advised to avoid the use entirely of